Paradroid STE DMA issue blog

[mst]

And just for anyones information - masteries STe also doesn't work on write either...

https://www.atari-forum.com/viewtopic.php?f=15&t=40000

I am begging to think anyone with an STe should just avoid ASCI.

I don't have a scope right now, I'm hoping geogeo can have a look so I pinged him.

Now this problem is fixed!!!

At least on my STE, but probably useful on another ST family machines.

Fix is rare, but interesting, also it has sense... today I focused on making the signal management faster, now the ACSI2STM manages the signal from the Atari a bit quicker, and it works now!

I spent all this afternoon, day 5 of February of 2021, playing the game SwitchBlade using savestate feature.
Writing 525188 Bytes to the hard disk each time I press the / key. Pressing * key, the game exits to desktop, and you can resume your game loading the created .TOS files with your savestate.

It worked more than 3 times in a row,

As a conclusion, the timing of signal is very restrictive, if you can´t deal with less than 200 ns signal management, your hard disk will miss valuable data.

This is the update firmware:

update.zip

Firmware update for the ACSI2STM in order to deal with strict ACSI timing on my STE. Writing to hard disk works ok now!

(25.93 KiB) Downloaded 457 times

[Steve]

Thank you @mst - this makes me wonder if the Ultrasatan could also be fixed for STE machines. I know Jookie states it is absolutely not a problem with the US, but after your discovery perhaps there is an improvement he could make. I'll get my ASCI2STM flashed asap and test

[geogeo_28]

No jazz...

Infact a little worse as it corrupted in both 32mhz and 8mhz modes. But still an interesting result, showing that the ASCI device can influence the result.

PXL_20210204_133052831.jpg

Just thinking out loud. As @exxos said it makes no sense for the Ultrasatan to be working in 32mhz but not 8mhz since the booster operates in 8mhz mode during DMA transfers. But since a read/write request has to be initiated by the ASCI device, R/W has to be high to initiate write and low to read. I wonder if the properties of RW change slightly depending on whether the machine is running in 8/32mhz. I don't have a probe currently so perhaps @geogeo_28 could check since we have the same issue?

Sorry for the delay, busy week..

RW WD1772 (pin 2) at 8MHz

IMG_0698.jpeg (92.55 KiB) Viewed 7655 times

RW WD1772 (pin 2) at 32MHz

IMG_0700.jpeg (86.15 KiB) Viewed 7655 times

Will do on the DMA now (pin1).

[geogeo_28]

R/W of the DMA at 8 MHz:

IMG_0702.jpeg (86.82 KiB) Viewed 7653 times

R/W of the DMA at 32 MHz:

IMG_0703.jpeg (88.98 KiB) Viewed 7653 times

[geogeo_28]

I am ready to try any new firmware of the UltraSatan
Running out of idea on my side. ^^

[geogeo_28]

And finally A1:

At 8 MHz:

IMG_0709.jpeg (87.33 KiB) Viewed 7650 times

IMG_0711.jpeg (101.94 KiB) Viewed 7648 times

At 32 MHz:

IMG_0707.jpeg (87.97 KiB) Viewed 7650 times

IMG_0710.jpeg (93.17 KiB) Viewed 7648 times

Definitely at 32 MHz, the signal is much faster and seems cleaner.

We can even see the 'spike' around 2.5V at 8MHz, which does not appear at 32 MHz..

[exxos]

Hmmm. Try a 390pf on A1 on the DMA chip,, see what chaos that causes..

[Steve]

Thanks for that @geogeo_28

I'm just busy flashing my ASCI2STM devices with the new firmware @mst has created for us. As his instructions are for Windows and I use Linux, there were a few differences so here is a guide on flashing for Linux:
(I used a CH341A programmer - first time I've ever used it actually)

1. The USB programmer usually has it's driver already installed in the kernel, so no driver needed.
2. stm32flash is available in Linux, just download from your repos or build from github
3. The port you need is most likely /var/ttyUSB0
4. A quick test to make sure it is correct is:

Code: Select all

stm32flash -b 115200 /dev/ttyUSB0 

Which should show:

Code: Select all

Interface serial_posix: 115200 8E1
Version      : 0x22
Option 1     : 0x00
Option 2     : 0x00
Device ID    : 0x0410 (STM32F10xxx Medium-density)
- RAM        : Up to 20KiB  (512b reserved by bootloader)
- Flash      : Up to 128KiB (size first sector: 4x1024)
- Option RAM : 16b
- System RAM : 2KiB

5. If it _doesn't_ work, or complains it cannot open the port, run this:

Code: Select all

sudo chmod a+rw /dev/ttyUSB0

6. This should work now, go ahead with the flash:

Code: Select all

stm32flash -b 115200 -w update.bin -v -g 0x8000000 /dev/ttyUSB0

PXL_20210206_162941328.jpg (436.8 KiB) Viewed 7615 times

[mst]

Thanks @Steve

Original firmware, and modified one are available for donwload,

Modified version works pretty well for STE equipped with "Good DMA" chip, but not capable of write on the hard disk.

[geogeo_28]

Hmmm. Try a 390pf on A1 on the DMA chip,, see what chaos that causes..

No change.
The signal looks better but still issue to write data on UltraSatan at 8 MHz.

IMG_0717.jpeg (91.63 KiB) Viewed 7573 times